AI Labs Will Trap You or Replace You - Dillon Rolnick, Nous Research

By ai.engineer

Categories: AI, Tools

Summary

AI Labs face a trillion-dollar ROI crisis as model commoditization erodes margins—forcing them to either trap users through agent-layer lock-in, replace workforces entirely, or build AGI. Founders must choose open foundations over proprietary platforms to avoid being held hostage.

Key Takeaways

  1. Token costs are commoditizing: average token price drops while total spending surges, but infrastructure costs remain flat—eliminating the margin that once existed in performant models.
  2. The AI battleground has shifted to the agent layer, not base models. Labs will embed agents into businesses to create switching costs and stickiness—making it economically hard to remove them later.
  3. Within 2-3 years, 70% of business workforce output will be agentic and controlled by third parties. If that controller has adverse incentives, your workforce can be 'organized into a strike.'
  4. Labs will eventually bypass partnerships and solve lucrative problems (e.g., finance) themselves. The trillion-dollar bet isn't sustainable through licensing—it requires direct market capture.
  5. Build on open foundations (Hermes Agent, open-source models) instead of proprietary platforms. Proprietary lock-in affects future business strategy, processes, and organizational intelligence itself.

Related topics

Transcript Excerpt

[music] >> I'm Dylan Rolnick. I'm the CEO of News Research. And today we're going to talk about building on foundations and platforms that you can actually own. Whoop, there we go. All right. How do you make back a cent a billion-dollar infrastructure and research bet? Labs individually have spent hundreds of billions of dollars training and building ever more sophisticated intelligence system and the industry as a whole has spent trillions perhaps. The problem is though that the models have begun to commoditize. You guys in this room who are hopefully practitioners of AI have known this intuitively. You switch among models readily. You switch between Astra, Fable, Opus, Kimmy, whatever it may be without too much regard for who's on the back end or what it may be. You want to try the new m…

More from ai.engineer